Habitat use, movement and activity of two large‐bodied native riverine fishes in a regulated lowland weir pool

Abstract The construction of dams and weirs, and associated changes to hydrological and hydraulic (e.g., water level and velocity) characteristics of rivers is a key environmental threat for fish. These multiple stressors potentially can affect fish in a variety of ways, including by causing changes in their movement, habitat use and activity. Understanding how and why these changes occur can inform management efforts to ameliorate these threats. In this context, we used acoustic telemetry to examine habitat use, longitudinal movement and activity of two lowland river fishes, Murray cod Maccullochella peelii and golden perch Macquaria ambigua, in a weir pool environment in south‐eastern Australia. We compared our results to published studies on riverine populations to determine if their behaviours are similar (or not). We show that M. peelii and M. ambigua in a weir pool exhibited some similar behaviours to conspecific riverine populations, such as strong site fidelity and use of woody habitat for M. ambigua. However, some behaviours, such as large‐scale (tens–hundreds of kilometres) movements documented for riverine populations, were rarely observed. These differences potentially reflect flow regulation (e.g., stable water levels, loss of hydraulic cues) in the weir pool. The two species also exhibited contrasting responses to dissolved oxygen conditions in the weir pool, which may reflect differences in their life history. Overall, this study shows that although some aspects of these two native fishes' life history can continue despite flow regulation, other aspects may change in weir pools, potentially impacting on long‐term population persistence.


| INTRODUCTION
The construction of dams and weirs to store or divert water for human use has altered the condition of aquatic ecosystems worldwide (Dudgeon et al., 2006;Jo et al., 2019;Olden, 2016;Vörösmarty et al., 2010). Major alterations include reduced flow variability, conversion of lotic to lentic habitats, reduced habitat complexity and altered water quality (Bunn & Arthington, 2002). These changes impact ecological processes and have effects on a range of biota, including fish, invertebrates and plants. For instance, conversion of lotic habitat to lentic in rivers due to the impoundment of upstream waters in dams and weirs, and associated changes to flow variability, water velocity, level and turbulence have been associated with the elimination of pelagic spawning fishes such as burbot Lota lota (Linnaeus 1758) in the Great River Ouse, UK (Copp, 1990), extinctions of aquatic snails in the lower River Murray, Australia (Sheldon & Walker, 1997), and decreased wood decomposition in regulated Mediterranean rivers (Abril et al., 2015).
Flow regulation can affect fish in a variety of ways, across cellular, individual, population and community levels (Murchie et al., 2008).
However, many animals, including fish, alter their behaviour as an initial response to environmental change and this can be a strong determinant of fitness outcomes (Wong & Candolin, 2015). Consequently, behavioural responses of fish (e.g., habitat use, movement, activity) to flow regulation are particularly important. For example, some fish shift position during periods of flow alteration or alter their habitat use or feeding, and these behaviours have implications for growth, condition and reproduction (Alexandre et al., 2016;Brenden et al., 2006;Del Mar Torralva et al., 1997). Understanding these changes to behaviour is critical therefore to evaluate impacts of alterations due to regulation and to guide management actions.
In Australia, regulation of river systems with dams and weirs has been extensive. In the Murray-Darling Basin (MDB), Australia's largest drainage basin, more than 4000 licensed in-stream structures have been constructed to regulate river flow (Lintermans, 2007). Along the main river in the MDB, the Murray River, weirs have created a series of contiguous pools for at least 700 km in the lower reaches (Walker, 2006). This modification has been identified as a driver of severe declines of biota, such as extinctions of native riverine fish species, including trout cod Maccullochella macquariensis (Cuvier 1829) and Macquarie perch Macquaria australasica (Cuvier 1830), and declines in other species such as silver perch Bidyanus bidyanus (Mitchell 1838) in the lower Murray River (Mallen-Cooper & Zampatti, 2018;Wedderburn et al., 2017). Elsewhere in the MDB dramatic fish kills have occurred from low dissolved oxygen concentrations (hypoxia) during reduced flows, such as in Rices Weir on Broken Creek and the Menindee Weir on the Darling River, with the latter attracting global media attention (Normile, 2019;Stewardson & Skinner, 2018).

Murray cod Maccullochella peelii (Mitchell 1838) and golden perch
Macquaria ambigua (Richardson 1845) are two large native riverine fish species of recreational angling and conservation significance endemic to south-eastern Australia. Populations of both species have declined substantially in the MDB due to factors such as habitat degradation and loss, altered flow regimes, reduced water quality and barriers to movement (Koehn & Nicol, 2014;Leigh & Zampatti, 2013;Thiem et al., 2017;Wedderburn et al., 2017). M. peelii is listed as Threatened nationally and M. ambigua as Near Threatened in Victoria. The two species display distinct life history strategies and thus may behave differently in response to environmental changes associated with river flow regulation, such as loss of hydraulic cues, altered water quality or reduced habitat complexity in weir pools.
Studies on riverine populations have shown that M. ambigua spawn and migrate in response to increases in flow or water level (Koster et al., 2017;Reynolds, 1983) whereas M. peelii spawning and migration is not dependent on flow increases (Humphries et al., 1999;Koehn et al., 2009).
M. peelii has a greater sensitivity to hypoxia than M. ambigua (Small et al., 2014) and there is evidence that M. ambigua may be more likely to move away from areas affected by hypoxic conditions than M. peelii (King et al., 2012;Koster et al., 2014b;Leigh & Zampatti, 2013). Both species use woody habitat, but M. peelii use deeper habitats, with higher water velocities, than M. ambigua (Koehn & Nicol, 2014). Given that environmental conditions such as hydraulics, water quality or habitat might be altered in weir pools, it is important to understand if fishes exhibit similar behaviour in weirs compared to rivers.
The aim of this study was to quantify daily and seasonal habitat use patterns and the longitudinal movement and activity of M. peelii and M. ambigua in relation to environmental conditions in a weir pool environment. We used a fine-scale continuous acoustic telemetry array, integrated with side-scanning sonar, to assess instream habitat conditions to (a) characterize the behaviour of fish in weir pools and evaluate if this is similar to knowledge about the traits and behaviours of these species in riverine populations; and (b) evaluate whether patterns differ among these two species. Knowledge of aspects of life history such as habitat use, movement and activity in weir pools is needed to better understand potential impacts of habitat change and flow regulation on these two species, and to inform potential management interventions. This information will also potentially inform efforts to assess and mitigate future impacts on other riverine fishes experiencing the effects of flow regulation.

| Study area
The study was conducted in lower Broken Creek, south-eastern Australia Average daily discharge in Broken Creek is 230 ml (Stewardson & Skinner, 2018). Broken Creek was subject to extensive historical removal of fallen trees and dredging (Trueman, 2011). Under natural conditions Broken Creek flowed intermittently, predominantly during winter-spring, and during most summers contracted to a series of pools (Reich et al., 2010). For the past 100 years a series of predominantly contiguous low-level weirs along lower Broken Creek and diversion channels has maintained permanent water for irrigation (Reich et al., 2010). These weirs are similar to those found in many other regulated rivers, that is, low-level (<3-5 m), narrow (<50-100 m wide) and shallow (<several metres deep) weirs operated to maintain a steady upstream pool level for irrigation (Baumgartner, 2007;Baumgartner et al., 2014;O'Connor et al., 2006). All of the weirs along lower Broken Creek have fishways installed capable of facilitating fish passage (O'Connor et al., 2006). Rices Weir, the most downstream weir on Broken Creek, exhibits water-quality issues, especially low dissolved oxygen concentrations (Stewardson & Skinner, 2018). Major fish kills have become a serious problem in lower Broken Creek in recent years, particularly at Rices Weir. The most likely explanation for the fish kills is low dissolved oxygen levels in conjunction with little or no flow (Rees, 2006).
Vemco V9 accelerometer transmitters (43 × 9 mm, frequency 69 kHz, mass 6.1 g in air, estimated battery life 350 days) were used for M. ambigua. Vemco V9 or V13 accelerometer transmitters (42 × 13 mm, frequency 69 kHz, mass 12.2 g in air, estimated battery life 740 days) were used for M. peelii, depending on the size of the fish. For both transmitter types the average delay between transmission was 240 s. Acceleration was sampled at 5 Hz for 30 s. The tag calculates a value that represents the root mean square acceleration that results from the combination of the acceleration from each of the three axes averaged over time.
For transmitter implantation, fish were transferred into an aerated, 50 l holding container of river water (temperature 11-13 C) and individually anaesthetized (0.03 ml AQUI-S per litre water) (AQUI-S, Lower Hutt, New Zealand). Time to anaesthesia was about 6-9 min.

| Habitat mapping
Instream woody habitat (IWH) and stream depth were measured in  Generalized additive mixed models (Hastie & Tibshirani, 1995) were used to examine patterns of daily and seasonal habitat use, using the mgcv (Wood, 2017) package in R (www.R-project.org). Habitat variables examined were (a) IWH volume (m 3 of wood); (b) count of IWH masses; and (c) maximum stream depth (m). These variables were selected because they are likely important for M. peelii and M. ambigua (Crook et al., 2001;Koehn & Nicol, 2014). Cyclic cubic regression splines (Wood, 2017) were used to smooth the response to hour-of-day and day-of-year. Individual fish were treated as a random effect. Any autocorrelation was assumed to be autoregressive with order 1 (AR1). Mean values for each habitat variable per hour were calculated for each individual across the entire monitoring period, when transmitters were detected within the intensive listening station array (88% of the time).

| Longitudinal movement
Generalized additive mixed models were used to explore relationships  (Royston & Sauerbrei, 2007) were used to smooth the response to these variables. Individual fish were treated as a random effect. Again, autocorrelations were assumed to be AR1. The model used a log-normal distribution. Distance moved (km) was calculated as the difference in mean daily position from the current day to the previous day (t 1 − t 0 ), with position defined as the distance of the receiver from Rices Weir. Total linear ranges for each fish were estimated by determining the distance between the most upstream and downstream locations in Broken Creek.

| Activity
Generalized additive mixed models were used to examine relationships between environmental variables and activity (measured in terms of m s −2 ) of M. peelii and M. ambigua, using the mgcv package in R. Explanatory variables examined were (a) average hourly discharge; (b) change (t 1 − t 0 ) in hourly flow; (c) average hourly dissolved oxygen;   Figure 3a). They were also more likely to use areas with greater IWH volume and counts of IWH masses around February, and areas with lower counts of IWH masses around November (Figure 3b,c).
M. ambigua were also more likely to be in deeper areas during the day, particularly from around midday to early afternoon, although the ecological significance of this result is questionable given a very small effect size (Figure 4). The generalised additive mixed models indicated no significant patterns of daily and seasonal habitat use for M. peelii.

| Activity
M. peelii and M. ambigua activity tended to gradually increase from August through to a peak in November-December and then decline or plateau to February (Figure 7). Activity in both species was strongly influenced by previous activity and change in dissolved oxygen (  (Crook, 2004;Crook et al., 2001;Koehn & Nicol, 2016). In comparison, some behaviours differed from those previously observed, especially the absence of longer-distance (tens to hundreds of kilometres) movements and a lack of habitat associations for M. peelii. The results also demonstrate interspecific variability in behaviours, such as disparate responses to dissolved oxygen conditions. These differences highlight the importance of understanding species-specific fish behaviour in response to changes in environmental conditions. Overall, the results suggest that some aspects of these two native fishes' life history persist despite flow regulation, but other aspects such as longer-distance movements may change in weir pools.

| Habitat use
We found that habitat use of M. ambigua was seasonally variable, with fish more likely to be in areas with greater IWH in February (austral late summer), and lower IWH in November. Seasonal changes in use of woody debris habitats are well known for many fishes, for example in brown trout Salmo trutta Linnaeus 1758, European eel Anguilla anguilla (Linnaeus 1758) and flathead catfish Pylodictis olivaris (Rafinesque 1818) (Weller & Winter, 2001). While M. ambigua were associated with woody debris in a short-term study (Crook et al., 2001) and can exhibit seasonality in habitat use (Koehn & Nicol, 2014), to our knowledge this is the first study to examine if associations with IWH vary seasonally. Woody debris can provide a range of habitat functions, including shelter, foraging sites and refuges from predators and high velocities (Crook et al., 2001). While the mechanism for the pattern we observed is unknown, one possible explanation could relate to spawning behaviour. Given M. ambigua are pelagic spawners, they may be less likely to inhabit areas with greater IWH masses during November, which is their peak spawning period in this region (Koster et al., 2014b(Koster et al., , 2017. Further work is needed to explore if this is the case, and more generally to improve our understanding of the spawning dynamics of fish that use weir pools. M. ambigua were also more likely to be in deeper habitats in August-December (late winter to early summer) and shallower habitats in January-February (mid to late summer). Seasonal movements between deeper and shallower habitats have also been reported for other fish species such as the congeneric Macquarie perch M. australasica Cuvier 1830 (Thiem et al., 2013) and the northern hog sucker Hypentelium nigricans (Lesueur 1817) (Matheney & Rabeni, 1995;Thiem et al., 2013). One possibility for the result observed here is that M. ambigua responds to changes in availability of prey.
We found no significant patterns of habitat use over time for M. peelii. This species is often associated with woody debris and deeper habitats (Jones & Stuart, 2007;Koehn, 2009;Koehn & Nicol, 2014). Seasonal differences in habitat use of M. peelii have also been reported, possibly related to increased water velocities, river widths and depths during higher flow periods (Koehn & Nicol, 2014). It is possible that differences between our study and these earlier ones may explain why different results were observed, such as the spatial scale at which they were conducted, the sampling locations and the methods that were used. Scales of observation, for example, are well recognized as potentially confounding estimates of habitat use (Crook et al., 2001;Hale et al., 2019;Koster & Crook, 2008). However, there are also plausible biological explanations why we may not have observed relationships, especially changes in the instream characteristics of lower Broken Creek associated with regulation, such as reduced flow variability and water level, as well as historical removal of habitat (i.e., fallen trees) and dredging.

| Longitudinal movement
We found that M. ambigua exhibited low levels of vagility, typically moving less than 3 km from their tagging location. Other studies have reported extended periods of restricted movement for this species (Crook, 2004;Zampatti et al., 2018), but these are often interspersed by periods when many individuals undertake large-scale movements (e.
g., tens to hundreds of kilometres), often in association with high flows (Koster et al., 2017;O'Connor et al., 2005). For example, a radio-tagging study of M. ambigua in the Murray River revealed that during late spring coinciding with increasing river discharge, 15 out of 19 (79%) tagged fish undertook long-distance (>10 km) movements (O'Connor et al., 2005). In the current study, there were several increases in discharge (e. g., from 185 to 449 ml day −1 in September 2012, 65 to 281 ml day −1 in August 2013), but the water level remained relatively constant due to regulation associated with Rices Weir.
Hydraulic-based cues such as a rise in water level rather than discharge per se are increasingly recognized as important to stimulating life history processes of riverine fishes such as movement (Dudley & Platania, 2007;Rakowitz et al., 2008).  Reynolds (1983) in the lower Murray River, which comprises extensive sections of weir pools, found some evidence of extensive movement by M. ambigua, this was associated with a rise in water level due to major flooding. In comparison, the changes in water level observed here may have been insufficient to elicit this response.
From a conservation and management perspective, approaches such as weir pool lowering and raising to create lotic habitats upstream may be an important management strategy for M. ambigua and other riverine fishes that rely on flow cues (Ye et al., 2008).
We found that M. peelii movements occurred over limited spatial extents, and most often during spring (September-November).
Movements were characterized by shifts away from usual locations, followed by return movements within 1-2 days, or shifts to new areas for 4-6 weeks where movement was again very limited, followed by return movements to areas previously occupied. Movements by M. peelii from a home location to a new position upstream, followed by a return movement to the area previously occupied during spring, have been observed in riverine and lake populations of M. peelii (the latter involving movements upstream into inflowing rivers) and may be spawning-related Koehn & Nicol, 2016). We found that these movements were undertaken by fish larger than (or just below) 500 mm in length, the approximate size of maturity for M. peelii (Rowland, 1998). The similarities in the spawning season movements of M. peelii from contrasting habitats (i.e., river, lake, weir pool) suggest that some aspects of life history can continue to be supported despite flow regulation.
The mating system of M. peelii involves nest-guarding and cleaning by male fish, while female fish leave the site once their eggs are deposited (Rowland, 1998). Although the sex of the fish could not be determined in our study, the return movements we observed within 1-2 days could be females visiting nest sites or males looking for mates or nesting sites. It is also possible that the shifts to new areas for 4-6 weeks could represent male nesting behaviour. Sexspecific variation in movement linked to reproductive behaviour has also been reported for other fish species such as freshwater catfish

| Activity patterns
M. ambigua were more active during periods of decreased dissolved oxygen. Fish respond to low dissolved oxygen in complex and varied ways, with increased activity potentially indicative of an avoidance response or increased swimming speed to improve gill ventilation (Chapman & McKenzie, 2009). While aquatic surface respiration has been documented for M. ambigua in response to hypoxia in laboratory conditions (Small et al., 2014), knowledge of the mechanisms M. ambigua use in the wild to cope with poor oxygen conditions is lacking. Determining whether the increased activity observed here relates to behavioural-physiological reactions such as increased gill ventilation or movements to avoid areas with low oxygen conditions would be valuable areas for future research. Indeed, the results of a study investigating the short-term effects of a blackwater (i.e., deoxygenation) event on fish, whereby more M. ambigua were collected in a non-affected site compared to a blackwater affected site (King et al., 2012), provide support for a hypothesis of movement of this species away (i.e., avoidance) from poor oxygen conditions.
In contrast to M. ambigua, M. peelii were less active during periods of decreased dissolved oxygen. In large fish such as M. peelii (up to 113.6 kg) (Lintermans, 2007) with high oxygen demands, decreased activity may represent an energy-saving strategy to allow them to cope with low levels of dissolved oxygen (Crocker & Cech, 1997). It has been shown under laboratory conditions that M. peelii is highly sensitive to hypoxia with a swimming capacity suited to energy conservation (Small et al., 2014;Whiterod, 2013). These characteristics may also therefore explain why M. peelii, which primarily inhabits lotic riverine environments less prone to hypoxia than lentic habitats, is often disproportionately reported in hypoxic blackwater fish kills (e.g., less likely to escape the hypoxic zone) (King et al., 2012;Small et al., 2014).
Understanding the resultant energetic expenditure for fish when they are exposed to particular water quality conditions such as low dissolved oxygen would be a useful next step for evaluation of biological consequences (Payne et al., 2011;Thiem et al., 2018). Nonetheless, our findings demonstrate the potential consequences of altered water quality as a result of modifications such as weir pools and the importance of water quality considerations in the management of flow regimes, which often focus heavily on water volume (Poff, 2018). Importantly, when dissolved oxygen decreased to about 3 mg l −1 , tagged M. peelii and M. ambigua were detected actively moving afterwards, demonstrating tolerance to short-term oxygen depletion. However, if fish are exposed to longer periods of dissolved oxygen depletion, the risk of mortality may be increased (Herbert et al., 2011;Kramer, 1987).
We found no meaningful effect of temperature or discharge on M. peelii and M. ambigua activity. In contrast, a study of M. peelii in the Edwards-Wakool river system showed that both variables influenced activity of this species (Thiem et al., 2018). The discrepancy between the two studies could relate to differences in methodology and instream characteristics of the study systems, such as hydraulic conditions, for example lack of variation in water levels and velocity in Broken Creek.
More generally, activity patterns of fishes can vary among different populations and even individuals of the same population (Metcalfe et al., 1999;Valdimarsson et al., 2000). There can be a variety of causes for this variability, including habitat, competition, and food availability and quality (Metcalfe et al., 1999;Metcalfe & Steele, 2001).
In conclusion, we found that some behaviours of the two focal species were similar in a weir pool to those observed in riverine populations but that there were also some considerable differences.
These results highlight the potential effects of environmental change